Loyalty Analytics
MAF Fall 2018 MeetingSeptember 21, 2018
PwC
Loyalty IntroductionBreakage / Redemption Rates
37
Modeling Breakage 10Attrition Risk 17Customer Lifetime Value 21Other Analytics 24
Agenda
PwC
Loyalty Introduction
PwC
Earning and redemption cycle gives rise to future redemptions on points already earned
Parallels with traditional insurance reserving
Loyalty Introduction
HospitalityHospitalityRetailRetail
AirlineAirline GamingGaming
Unpaid Claim
Reserve
Unpaid Claim
ReserveLoyalty LiabilityLoyalty Liability
3.8 billion loyalty program memberships in US in 2016*
Loyalty programs span many industries
* 2017 Colloquy Loyalty Census
PwC
Re-design loyalty programs to ensure that the finances are sound
Revise policy language / forms
Valuing Points
Currency Actuarial Pricing
Other Parallels Between Loyalty Programs & Insurance
Program Design Underwriting &
Policy Form Updates
Determine or revise the pricing of points for sale to external parties (e.g., credit card companies or points.com)
Ratemaking
PwC
• The redemption rate % is generally the harder input to estimate with accuracy
• Traditional triangular methods often used
• Redemption CPP has a shorter tail and changes manifest more rapidly
Loyalty Liability Estimates
Estimated Redemption
Rate%
# of Points Outstanding
Redemption Cost per Point
$
Rewards Liability
$
Future Points to be Redeemed
PwC
Breakage / Redemption Rates
PwC
Earn & Burn
Customer earns benefits while
interacting with company
Customer redeems earned
benefits
Benefits remain unused
Customer halts interactions
and/or expiration
PwC
Usage & Breakage Rates
Usage rate – Percentage of points earned/outstanding which will ultimately be redeemed
Breakage rate – Percentage of points earned/outstanding which will not ultimately be redeemed
100%
12%
88%
31%
69%
48%
52%
68%
32%
Unused
Redeemed
PwC
Modeling Breakage
PwC
A Simple SolutionA
ctua
l / P
redi
cted
U
ltim
ate
Usa
ge R
ate
Predictor
GLM fitted to historically observed usage rates
~
PwC
A Simple SolutionA
ctua
l / P
redi
cted
U
ltim
ate
Usa
ge R
ate
Predictor
GLM fitted to historically observed usage rates
~ Future usage predicted using fitted
coefficients
∗
x*
PwC
A (Not So) Simple Solution
GLM fitted to predict 12m usage
GLM fitted to predict
24m usage
GLM fitted to predict
36m usage
GLM fitted to predict
ultimate usage………
PwC
A (Not So) Simple Solution
…
PwC
A (Not So) Simple Solution
0 0.5 1 1.5 2 2.50%
10%
20%
30%
40%
50%
60%
70%
80%
Years
Red
empt
ion
Rat
e
Ult
PwC
0 0.5 1 1.5 2 2.50%
10%
20%
30%
40%
50%
60%
70%
80%
Years
Red
empt
ion
Rat
e
Ult
A (Not So) Simple Solution
PwC
Attrition Risk
PwC
Attrition Risk Scores
Attrition Risk – Likelihood that customer will stop interacting with the company by some future date
- Insurance – Easy to define an attrition event
- Loyalty – Need to define an objective measure
Attrition Risk Score – Measure of likelihood of attrition, possibly banded
Considerations- Future period of time
- Significant reduction in activity
- Earn & Burn
Find sub-population characteristics with a high correlation with attrition risk
PwC
Dynamic Attrition Risk Scoring
0%
20%
40%
60%
80%
100%
1‐Jan‐17 1‐Feb‐17 1‐Mar‐17 1‐Apr‐17 1‐May‐17 1‐Jun‐17
AttritionRiskScore
EarlyWarning MemberA MemberB
PwC
Customer Lifetime Value
PwC
Customer Lifetime Value
The value a customer brings to a loyalty program, expressed as a lifetime dollar value.
PwC
Components of Value
CLV
Activity period
Revenue generated
Cost of Acquisition
Cost of Retention
Cost of capital
Time value of money
PwC
Other Analytics
Cost per Point New Business Conversion
Tier Movements Customer Segmentation
Advanced Analytics
PwC
Loyalty Analytics: Measuring Loyalty Program Performance Whitepaper
Accessed via: http://pwc.to/2tu6y1C
1. Changing the loyalty industry
2. Identifying loyalty program goals
3. Developing analytic targets
4. Loyalty models and metrics
5. Taking action
PwC
Questions?
Jean-François Greeff – Manager, PwC Actuarial [email protected]
Mark Doucette – Director, PwC Actuarial [email protected]